Optimal Estimation For Networked Systems With Communication Constraints

In recent years,with the rapid development of network technology,state estimation for networked system has attracted increasing attention,and finds applications in a variety of areas,such as target tracking,fault detection and intelligent transportation.The introduction of communication network in estimation systems has brought many advantage,meanwhile,it has also induced several new problems and difficulties for state estimation.Based on the theory of optimal estimation and optimal weighed fusion methods,this thesis is concerned with two important problems in networked estimation systems with communication constraints,one is the asynchronous sampling and transmitting,and the other is the missing measurements.The main work of the thesis is summarized as follows:1.Centralized estimation algorithm and sequential estimation algorithm are proposed to deal with asynchronous sampling and transmitting problem for the networked estimation system.In centralized estimation algorithm,the measurements are augmented to perform the filtering task after all the sensors transmitted the measurements to the remote estimator.The sequential estimator performs the filtering task according to the order of measurements transmitted by the sensors.Both the two methods are extended to the case of packet dropouts,and the retransmission scheme is also considered.2.For the networked system with communication constraints,a novel compensation strategy is presented to compensate the components in remote estimator which are not updated.Bernoulli distribution is applied to describe the accessing process of each sensor node.By using orthogonal projection principle,a multi-sensor-based optimal estimator is designed for the networked systems with media access constraints.An illustrative example is presented to show the effectiveness of the proposed methods.3.A fusion estimation problem is considered in the multi-sensor networked system with communication constraints.Centralized measurement fusion estimator and weighted measurement fusion estimator are designed based on measurement augmentation and weighted least squares method.Moreover,the functional equivalence of the two algorithms is proved under such a constraint.Finally,a continuous stirred tank reactor is provided to demonstrate theeffectiveness of the proposed methods.Conclusions and future work are presented in the end of the thesis.